Methods and systems may provide for cognitive decline prediction or assessment. Baseline and follow-up inertial sensor data may be received from one or more inertial sensors attached to a person. Baseline and follow-up data indicative of cognitive decline may be received from the person. An indication of cognitive decline may be determined based on the baseline and follow-up cognitive decline data. A classifier function for predicting cognitive decline may be trained with the baseline inertial sensor data and the indication of cognitive decline. A classifier function for assessing cognitive decline may be trained with the baseline inertial sensor data, a difference between the baseline inertial sensor data and follow-up inertial sensor data, and the indication of cognitive decline.
Legal claims defining the scope of protection. Each claim is shown in both the original legal language and a plain English translation.
1. A computer-readable non-transitory medium comprising computer-readable code physically embodied thereon which, when executed by a processor, causes the processor to perform a method comprising: receiving inertial sensor data directly from one or more inertial sensors attached to a person's shank that performs a quantitative timed-up-and-go (QTUG) test, the inertial sensor data being measured by the one or more inertial sensors and including shank angular velocity data comprising measured gait parameters related to the person sitting, walking, and turning; receiving, at a time subsequent to receiving the inertial sensor data from the QTUG test, additional inertial sensor data indicative of cognitive function of the person directly from the one or more inertial sensors, including shank angular velocity data comprising gait parameters related to the person sitting, walking, and turning measured during a subsequent QTUG test; determining an indication of cognitive decline in the person based on a difference between the inertial sensor data indicative of cognitive function and the received additional inertial sensor data including a difference between the gait parameters; generating a classifier function that associates the received inertial sensor data with the determined indication of cognitive decline in the person as a prediction for cognitive decline in other persons; after generation of the classifier function, collecting inertial sensor data directly from the one or more inertial sensors from another person that performs the QTUG test; using the generated classifier function, associating the collected inertial sensor data from the QTUG test from the another person to an indication of cognitive decline in the another person; and providing, using the association from the generated classifier function, to a user a prediction of whether the another person will experience cognitive decline based on the inertial sensor data collected from the another person, wherein the prediction provided to the user comprises a classification of the another person as either having cognitive decline or being cognitively intact.
A computer program stored on a non-transitory medium analyzes movement data from inertial sensors (like accelerometers and gyroscopes) attached to a person's lower leg (shank) during a Timed Up and Go (TUG) test to predict cognitive decline. The program receives initial sensor data (shank angular velocity data related to sitting, walking, and turning), then receives follow-up sensor data from a later TUG test. It determines cognitive decline by comparing the initial and follow-up movement data, specifically changes in gait parameters. It then creates a classifier function, which is a mathematical model that links movement data to cognitive decline. Finally, it uses this model to predict cognitive decline in new individuals based on their TUG test data, classifying them as either having cognitive decline or being cognitively intact, and provides this prediction to a user.
2. The computer-readable non-transitory medium of claim 1 , wherein the receiving the data indicative of cognitive decline in the person occurs one year or longer after the receiving inertial sensor data.
The computer program from the previous description where inertial sensor data is analyzed from a TUG test to predict cognitive decline, specifies that the follow-up inertial sensor data is collected one year or longer after the initial inertial sensor data. This delayed collection allows for observation of longer-term changes in gait that may be indicative of cognitive decline.
3. The computer-readable non-transitory medium of claim 2 , wherein the generating the classifier function further comprises calculating a statistical parameter of the shank angular velocity data, and wherein the classifier function is further based on the statistical parameter of the shank angular velocity.
The computer program from the previous cognitive decline prediction description, where follow-up data is collected a year or more after baseline, refines the classifier function by calculating statistical parameters (e.g., average, variance) from the shank angular velocity data. These statistical parameters, in addition to the raw sensor data, are used to generate the classifier function. This provides a more robust model by considering aggregate measures of movement rather than just instantaneous values.
4. The computer-readable non-transitory medium of claim 3 , wherein the inertial sensor data and additional inertial sensor data comprises data indicative of a walk ratio or a turn-end time during the QTUG test, the walk ratio being a ratio of (a) a time taken to stand and walk to a median turning point and (b) a time taken to walk back from the median turning point and sit to complete the test and the turn-end time being a time taken during the QTUG test to walk back from the median turning point and sit to complete the test.
The computer program from the previous description that generates a cognitive decline classifier function based on statistical parameters, specifically uses data indicative of a "walk ratio" (time to stand and walk to midpoint vs. time to walk back and sit) and "turn-end time" (time to walk back from the midpoint and sit) during the TUG test. These parameters provide specific insights into gait and turning performance, which are sensitive to cognitive function.
5. The computer-readable non-transitory medium of claim 4 , further comprising receiving an age of the person, wherein the generating the classifier function is further based on the age in addition to the received inertial sensor data and the determined indication of cognitive decline in the person and wherein the classifier function is configured to predict whether the another person will experience cognitive decline based on the another person's age and the collected inertial sensor data.
The computer program from the previous description that uses walk ratio and turn-end time in a cognitive decline prediction model, also incorporates the person's age. The classifier function is generated using age, inertial sensor data, and cognitive decline indication. This allows the model to account for age-related changes in gait that are independent of cognitive decline and provides more accurate predictions tailored to different age groups. The model then predicts cognitive decline in other persons based on their age and collected inertial sensor data.
6. The computer-readable non-transitory medium of claim 5 , further comprising receiving grip strength data of the person, wherein generating the classifier function is further based on the person's grip strength data, and wherein the classifier function is configured to predict whether the another person will experience cognitive decline further based on the another person's grip strength.
The computer program from the previous description that uses age and movement data to predict cognitive decline, further incorporates the person's grip strength. The classifier function is generated based on grip strength, age, inertial sensor data and indication of cognitive decline. The classifier function is also configured to predict cognitive decline based on a new person's grip strength, in addition to the new person's age and movement data, to further improve accuracy.
7. The computer-readable non-transitory medium of claim 1 , wherein the received data indicative of cognitive function in the person comprises a follow-up mini-mental state exam (MMSE) score, the method further comprising: receiving a baseline MMSE score before receiving the follow-up MMSE score, wherein the determining the indication of cognitive decline in the person comprises determining a difference between the baseline MMSE score and the follow-up MMSE score, the difference between the scores being used to create the classifications in the generated classifier function of cognitive decline and cognitively intact.
The computer program from the initial cognitive decline prediction description that analyzes inertial sensor data from a TUG test uses a Mini-Mental State Exam (MMSE) score to determine cognitive decline. Specifically, it uses a follow-up MMSE score and compares it to a baseline MMSE score taken before the follow-up. The difference between these MMSE scores is then used to create classifications (cognitive decline vs. cognitively intact) that are used to train the classifier function.
8. A computer-readable non-transitory medium comprising computer-readable code physically embodied thereon which, when executed by a processor, causes the processor to perform a method comprising: receiving baseline inertial sensor data directly from one or more inertial sensors attached to a person's shank that performs a quantitative timed-up-and-go (QTUG) test, the received baseline inertial sensor data being measured by the one or more inertial sensors and comprising shank angular velocity data related to the person sitting, walking, and turning; receiving, at a time subsequent to the receiving baseline inertial sensor data, follow-up inertial sensor data from the person that performs the QTUG test directly from the one or more inertial sensors, the follow-up inertial sensor data comprising shank angular velocity data related to the person sitting, walking, and turning measured during a subsequent QTUG test; determining a change between the baseline inertial sensor data and the follow-up inertial sensor data; determining an indication of cognitive decline in the person based on the baseline inertial sensor data; determining an indication of cognitive decline in the person based on the change; generating a classifier function that associates the received baseline inertial sensor data and the determined change with the indication of cognitive decline in the person; then after generation of the classifier function, receiving other baseline inertial sensor data from another person that performs the QTUG test directly from the one or more inertial sensors; receiving other follow-up inertial sensor data directly from the one or more inertial sensors from the another person that performs the QTUG test thereafter; determining a change between the other baseline inertial sensor data and the follow-up inertial sensor data of the another person; using the generated classifier function, associating the follow-up inertial sensor data and the determined change from the another person to an indication of cognitive decline in the another person; and providing, using the association of the generated classifier function, to a user an assessment of whether the another person has experienced cognitive decline based on the other baseline inertial sensor data and the other follow-up inertial sensor data collected from the another person, wherein the assessment provided to the user comprises a classification of the another person as either having cognitive decline or being cognitively intact.
A computer program stored on a non-transitory medium analyzes movement data from inertial sensors attached to a person's lower leg during a TUG test to assess cognitive decline. The program receives baseline sensor data, and later receives follow-up sensor data from a subsequent TUG test. It determines the change between baseline and follow-up sensor data and determines an indication of cognitive decline based on both the baseline data and the change. It then generates a classifier function to associate the baseline data and the change with the indication of cognitive decline. Finally, it uses the generated classifier function to provide an assessment of whether another person has experienced cognitive decline based on their baseline and follow-up data, classifying them as having cognitive decline or being cognitively intact.
9. The computer-readable non-transitory medium of claim 8 , wherein the method further comprises receiving a follow-up mini-mental state exam (MMSE) score, the method further comprising receiving a baseline MMSE score before receiving the follow-up MMSE score, wherein the determining the indication of cognitive decline in the person further comprises determining a difference between the baseline MMSE score and the follow-up MMSE score, the difference between the scores being used to create the classifications in the generated classifier function of cognitive decline and cognitively intact.
The computer program from the previous cognitive decline assessment description that analyzes inertial sensor data from a TUG test, also receives and compares Mini-Mental State Exam (MMSE) scores. It compares a follow-up MMSE score to a baseline MMSE score, where the baseline score was taken before the follow-up. The difference between these scores is used to classify individuals as either having cognitive decline or being cognitively intact, which is used in the classifier function.
10. The computer-readable non-transitory medium of claim 8 , wherein the baseline inertial sensor data comprises a baseline X-axis shank angular velocity data or baseline Z-axis shank angular velocity data, wherein the follow-up inertial sensor data comprises a follow-up X-axis shank angular velocity data or follow-up Z-axis shank angular velocity data, wherein the determining the change comprises determining a change between a statistical parameter of the baseline X-axis shank angular velocity data or of the baseline Z-axis shank angular velocity data and a statistical parameter of the follow-up X-axis shank angular velocity data or of the follow-up Z-axis shank angular velocity data.
The computer program from the previous description that uses TUG inertial sensor data to assess cognitive decline, specifically analyzes X-axis and Z-axis shank angular velocity data. It calculates a change between statistical parameters of the baseline X or Z-axis data and the follow-up X or Z-axis data. This allows for detecting changes in specific movement patterns related to cognitive function.
11. The computer-readable non-transitory medium of claim 10 , wherein the baseline inertial sensor data further comprises a mean single support percentage, wherein the follow-up inertial sensor data comprises a follow-up single support percentage, and wherein the determining the change comprises determining a change between the baseline single support percentage and the follow-up single support percentage.
The computer program from the previous description that uses X/Z-axis shank angular velocity statistical parameter changes in cognitive decline assessment, also incorporates "single support percentage" data. It measures the change between baseline single support percentage and follow-up single support percentage, which is indicative of balance and gait stability.
12. The computer-readable non-transitory medium of claim 11 , wherein the classifier function is a regularized discriminant classifier function.
The computer program from the previous description that utilizes single support percentage change for cognitive decline assessment, specifies that the classifier function is a "regularized discriminant classifier function". This type of classifier function is a specific type of machine learning model that helps to prevent overfitting to the training data and improve generalization to new data.
13. The computer-readable non-transitory medium of claim 8 , wherein the receiving the follow-up data from the person occurs one year or longer after the receiving baseline inertial sensor data.
The computer program from the initial cognitive decline assessment description that analyzes inertial sensor data from a TUG test specifies that the follow-up data from the person is collected one year or longer after receiving the baseline inertial sensor data.
14. A system for predicting a cognitive decline in a person, the system comprising: one or more inertial sensors attached to a person's shank and in communication with a processor; and a non-transitory storage medium operatively coupled to the processor, the storage medium having instructions disposed thereon, which when executed by the processor, cause the processor to: receive inertial sensor data directly from one or more inertial sensors attached to a person that performs a quantitative timed-up-and-go (QTUG) test, including shank angular velocity data comprising measured gait parameters related to the person sitting, walking, and turning, receive, at a time subsequent to receiving the inertial sensor data from the QTUG test, additional inertial sensor data directly from the one or more inertial sensors including shank angular velocity data comprising gait parameters related to the person sitting, walking, and turning measured during a subsequent QTUG test, determine an indication of cognitive decline in the person based on a difference between the received inertial sensor data and the additional inertial sensor data, generate a classifier function that associates the received inertial sensor data with the indication of cognitive decline in the person; after generation of the classifier function, collect inertial sensor data directly from the one or more inertial sensors from another person that performs the QTUG test; use the generated classifier function to associate the collected inertial sensor data from the QTUG test from the another person to an indication of cognitive decline in the another person, and provide, using the association from the generated classifier function, to a user a prediction of whether the another person will experience cognitive decline based on the inertial sensor data collected from the another person, wherein the prediction provided to the user comprises a classification of the another person as either having cognitive decline or being cognitively intact.
A system for predicting cognitive decline includes inertial sensors attached to a person's shank and a processor. The system receives inertial sensor data from a TUG test (shank angular velocity related to sitting, walking, and turning), then receives follow-up sensor data. It determines cognitive decline based on a difference between the initial and follow-up data, generates a classifier function to link the data to cognitive decline, and uses the classifier to predict decline in other people based on their TUG data. The system provides a prediction (cognitive decline or cognitively intact) to a user.
15. The system of claim 14 , wherein the additional inertial sensor data received by the processor occurs after a time period that is one year or longer after receiving inertial sensor data.
The cognitive decline prediction system described previously where inertial sensor data is analyzed from a TUG test requires the processor to receive the additional inertial sensor data (follow-up) after a time period of one year or longer after receiving the initial inertial sensor data.
16. A system for predicting a cognitive decline in a person, the system comprising: one or more inertial sensors attached to a person's shank that performs a quantitative timed-up-and-go (QTUG) test and that are in communication with a processor; and a non-transitory storage medium operatively coupled to the processor, the storage medium having instructions disposed thereon, which when executed by the processor, cause the processor to: receive baseline inertial sensor data directly from the one or more inertial sensors, the received baseline inertial sensor data being measured by the one or more inertial sensors and comprising shank angular velocity data related to the person sitting, walking, and turning during the QTUG test, receive directly from the one or more inertial sensors, at a time subsequent to receiving the baseline inertial sensor data, follow-up inertial sensor data from the person, the follow-up inertial sensor data comprising shank angular velocity data related to the person sitting, walking, and turning measured during a subsequent QTUG test, determine a change between the baseline inertial sensor data and the follow-up inertial sensor data, determine an indication of cognitive decline in the person based on the baseline inertial sensor data, determine an indication of cognitive decline in the person based on the change; generate a classifier function that associates the received baseline inertial sensor data and the determined change in inertial sensor data with the indication of cognitive decline in the person, receive other baseline inertial sensor data directly from the one or more inertial sensors from another person that performs the QTUG test; receive other follow-up inertial sensor data directly from the one or more inertial sensors from the another person that performs the QTUG test thereafter, and provide, using the association from the generated classifier function, to a user an assessment of whether the another person has experienced cognitive decline based on the other baseline inertial sensor data and the other follow-up inertial sensor data collected from the another person, wherein the assessment provided to the user comprises a classification of the another person as either having cognitive decline or being cognitively intact.
A system for assessing cognitive decline includes inertial sensors on a person's shank and a processor. The system receives baseline sensor data from a TUG test, then receives follow-up sensor data. It determines the change between baseline and follow-up data, and determines an indication of cognitive decline based on both. A classifier function is generated to link baseline data and changes to cognitive decline. The system receives data from another person, calculates the change, uses the classifier function to assess their cognitive decline, and provides an assessment to a user (cognitive decline or cognitively intact).
17. The system of claim 16 , wherein the follow-up inertial sensor data received by the processor occurs after a time period that is one year or longer after receiving baseline inertial sensor data.
The cognitive decline assessment system described previously where inertial sensor data is analyzed from a TUG test requires the processor to receive the follow-up inertial sensor data after a time period of one year or longer after receiving the baseline inertial sensor data.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
July 20, 2012
May 23, 2017
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.